RE: [rules-users] Drools 4 poor performance scaling?
hi ,
please provide more information on the above technique .
Let me put what i have understood from the above explanation .
u have around 700k - 900k total facts.
first u are inserting single fact ie (ColdStarting fact) and retracts it (
how to retract it ?) when all the Jobs
and Workers have been loaded.( it means all the other facts are loaded ??)
This changed our startup time from over 50 minutes to under 5. what time
it is ??
There's some sort of strange propagation and looping going on with
accumulation on the fly, at least with our facts and rules. please exaplin
more on this ?
please provide the url on the wiki where we get more info on this .
Thank you .
Fenderbosch, Eric wrote:
FYI for the group. We seem to have solved our performance problem.
I'll describe our problem space a bit some people have some context. We
load up about 1200 Jobs with about 3000 Stops and about 1500 Vehicles
with about 2000 Workers. We then calculate Scores for each Vehicle for
each Job. Some combinations get excluded for various reasons, but we
end up with 700k - 900k total facts. We do score totaling and sorting
using accumulators.
One of our teams members (nice find Dan) decided to try to isolate the
accumulation rules until all our other facts are loaded. Those rules
now have a not ColdStarting() condition and our startup code inserts a
ColdStarting fact as the first fact and retracts it when all the Jobs
and Workers have been loaded. This changed our startup time from over
50 minutes to under 5. There's some sort of strange propagation and
looping going on with accumulation on the fly, at least with our facts
and rules.
I'll put an entry on the wiki as well.
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Fenderbosch,
Eric
Sent: Monday, June 30, 2008 11:46 AM
To: Rules Users List
Subject: RE: [rules-users] Drools 4 poor performance scaling?
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
Re: [rules-users] Drools 4 poor performance scaling?
It's more related to the issues with accumulate.
At the moment ever single propagation executes both the action and the
result and propagates the result to the next pattern. When dataloading
this means you are constantly evaluating the result and the rest of the
rule for each propagation. When in realty what you want to do is load
the accumulate and execute the result and propagate at the end - however
trying to find a robust way of doing this isn't easy.
Mark
Anstis, Michael (M.) wrote:
I wonder whether is's a benefit of truth maintenance? If a new fact is
inserted into working memory that could cause an activation of a rule
that contains an accumulate (or collect) to change then the whole
accumulate (or collect) operator is executed again?!?
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Fenderbosch,
Eric
Sent: 09 July 2008 14:41
To: Rules Users List
Subject: RE: [rules-users] Drools 4 poor performance scaling?
FYI for the group. We seem to have solved our performance problem.
I'll describe our problem space a bit some people have some context. We
load up about 1200 Jobs with about 3000 Stops and about 1500 Vehicles
with about 2000 Workers. We then calculate Scores for each Vehicle for
each Job. Some combinations get excluded for various reasons, but we
end up with 700k - 900k total facts. We do score totaling and sorting
using accumulators.
One of our teams members (nice find Dan) decided to try to isolate the
accumulation rules until all our other facts are loaded. Those rules
now have a not ColdStarting() condition and our startup code inserts a
ColdStarting fact as the first fact and retracts it when all the Jobs
and Workers have been loaded. This changed our startup time from over
50 minutes to under 5. There's some sort of strange propagation and
looping going on with accumulation on the fly, at least with our facts
and rules.
I'll put an entry on the wiki as well.
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Fenderbosch,
Eric
Sent: Monday, June 30, 2008 11:46 AM
To: Rules Users List
Subject: RE: [rules-users] Drools 4 poor performance scaling?
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1
RE: [rules-users] Drools 4 poor performance scaling?
FYI for the group. We seem to have solved our performance problem.
I'll describe our problem space a bit some people have some context. We
load up about 1200 Jobs with about 3000 Stops and about 1500 Vehicles
with about 2000 Workers. We then calculate Scores for each Vehicle for
each Job. Some combinations get excluded for various reasons, but we
end up with 700k - 900k total facts. We do score totaling and sorting
using accumulators.
One of our teams members (nice find Dan) decided to try to isolate the
accumulation rules until all our other facts are loaded. Those rules
now have a not ColdStarting() condition and our startup code inserts a
ColdStarting fact as the first fact and retracts it when all the Jobs
and Workers have been loaded. This changed our startup time from over
50 minutes to under 5. There's some sort of strange propagation and
looping going on with accumulation on the fly, at least with our facts
and rules.
I'll put an entry on the wiki as well.
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Fenderbosch,
Eric
Sent: Monday, June 30, 2008 11:46 AM
To: Rules Users List
Subject: RE: [rules-users] Drools 4 poor performance scaling?
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
StatefulSession wm = ruleBase.newStatefulSession();
// Global - output
warnings = new Alarm();
wm.setGlobal(warnings, warnings);
// Add facts
st = (new Date()).getTime();
for(i=0; i nfacts; i++) {
wm.insert(new Data(rand.nextInt(4),
rand.nextDouble()-0.5));
}
en = (new Date()).getTime();
System.out.println(facts = + (en-st));
// Now run the rules
st = (new Date()).getTime();
wm.fireAllRules();
en = (new Date()).getTime();
System.out.println(rules = + (en-st));
// Clean up
wm.dispose();
System.out.println(\n);
}
This code is based on the HelloWorldExample.java code from
RE: [rules-users] Drools 4 poor performance scaling?
I wonder whether is's a benefit of truth maintenance? If a new fact is
inserted into working memory that could cause an activation of a rule
that contains an accumulate (or collect) to change then the whole
accumulate (or collect) operator is executed again?!?
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Fenderbosch,
Eric
Sent: 09 July 2008 14:41
To: Rules Users List
Subject: RE: [rules-users] Drools 4 poor performance scaling?
FYI for the group. We seem to have solved our performance problem.
I'll describe our problem space a bit some people have some context. We
load up about 1200 Jobs with about 3000 Stops and about 1500 Vehicles
with about 2000 Workers. We then calculate Scores for each Vehicle for
each Job. Some combinations get excluded for various reasons, but we
end up with 700k - 900k total facts. We do score totaling and sorting
using accumulators.
One of our teams members (nice find Dan) decided to try to isolate the
accumulation rules until all our other facts are loaded. Those rules
now have a not ColdStarting() condition and our startup code inserts a
ColdStarting fact as the first fact and retracts it when all the Jobs
and Workers have been loaded. This changed our startup time from over
50 minutes to under 5. There's some sort of strange propagation and
looping going on with accumulation on the fly, at least with our facts
and rules.
I'll put an entry on the wiki as well.
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Fenderbosch,
Eric
Sent: Monday, June 30, 2008 11:46 AM
To: Rules Users List
Subject: RE: [rules-users] Drools 4 poor performance scaling?
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
StatefulSession wm = ruleBase.newStatefulSession();
// Global - output
warnings = new Alarm();
wm.setGlobal(warnings, warnings);
// Add facts
st = (new Date()).getTime();
for(i=0; i nfacts; i++) {
wm.insert(new Data(rand.nextInt(4),
rand.nextDouble
RE: [rules-users] Drools 4 poor performance scaling?
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
StatefulSession wm = ruleBase.newStatefulSession();
// Global - output
warnings = new Alarm();
wm.setGlobal(warnings, warnings);
// Add facts
st = (new Date()).getTime();
for(i=0; i nfacts; i++) {
wm.insert(new Data(rand.nextInt(4),
rand.nextDouble()-0.5));
}
en = (new Date()).getTime();
System.out.println(facts = + (en-st));
// Now run the rules
st = (new Date()).getTime();
wm.fireAllRules();
en = (new Date()).getTime();
System.out.println(rules = + (en-st));
// Clean up
wm.dispose();
System.out.println(\n);
}
This code is based on the HelloWorldExample.java code from the manual
and the setup for the rule base is the same as in the manual. As the
number of facts increases runtime increases dramatically:
facts -- runtime (ms)
10 -- 168
20 -- 166
30 -- 344
40 -- 587
50 -- 1215
60 -- 1931
70 -- 2262
80 -- 3000
90 -- 4754
with a maximum memory use of about 428 MB RAM. By contrast, if I use
sequential stateless sessions, everything runs in about 1-5 ms.
Is there something in my set up that would cause this, or is this how
one would expect Drools to scale? I read about people using thousands
of facts so I suspect I'm setting something up incorrectly.
Any help appreciated!
Ron
_
The other season of giving begins 6/24/08. Check out the i'm Talkathon.
http://www.imtalkathon.com?source=TXT_EML_WLH_SeasonOfGiving
___
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
___
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
Re: [rules-users] Drools 4 poor performance scaling?
Fenderbosch, Eric wrote:
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
If you have multi-cpu there is a JVM command you can set a dedicated cpu
for GC, that helps somewhat.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
Other than the CPU thing, Drools won't take advantage of multipe cpus at
the moment.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
It could be related to this, especially if you create a long chain of
logical relationships.
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Have you tried this on Drools 5.0? It 'doesn't need shadow proxies and
implements a new Rete algorithm that is faster for retracts. You can get
a nightly build from here, I'd be interested to find out how broken 5.0
is :)
https://hudson.jboss.org/hudson/job/drools/lastSuccessfulBuild/artifact/trunk/target/
We still have more performnace work to do, the items are known, just a
matter of time, not all will make 5.0 though. but the main items include:
1) bytecode compiled Rete network, instead of interpreted nodes. I'm
hoping this will have a large impact, reducing GC and general
indirection and recursive method call frames.
2) true modify, instead of a retract+assert, will also remove the need
for activation normalistaion that we do for TMS and the agenda event model.
3) range indexing (initially literals, but would like to explore
variables too).
Steve, before he left fedex, was creating a simulator for this use case,
but removing anything business sensitive. So that we could use it
publicly as a benchmark and to help us tune the engine. Are you still
working on this? Steve use to chat to us on irc, can I ask you to pop on
for a chat?
http://labs.jboss.org/drools/irc.html
mark
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
StatefulSession wm = ruleBase.newStatefulSession();
// Global - output
warnings = new Alarm();
wm.setGlobal(warnings, warnings);
// Add facts
st = (new Date()).getTime();
for(i=0; i nfacts; i++) {
wm.insert(new Data(rand.nextInt(4),
rand.nextDouble()-0.5));
}
en = (new Date()).getTime();
System.out.println(facts = + (en-st));
// Now run the rules
RE: [rules-users] Drools 4 poor performance scaling?
Eric-
Mark has my test code and said he'll look at it in a few days. I'm sure he'll
reply here but if not, I'll post it. My question for you is how quickly does
fireAllRules() run for you? As I said, with only 95 facts loaded it is taking
my machine (dual core 2 GHz box, Fedora Core 6) almost 5 seconds.
Ron
Subject: RE: [rules-users] Drools 4 poor performance scaling?
Date: Mon, 30 Jun 2008 11:46:15 -0400
From: [EMAIL PROTECTED]
To: rules-users@lists.jboss.org
We are having a similar problem, although our fact count is much higher.
Performance seems pretty good and consistent until about 400k facts,
then performance degrades significantly. Part of the degradation is
from bigger and more frequent GCs, but not all of it.
Time to load first 100k facts: ~1 min
Time to load next 100k facts: ~1 min
Time to load next 100k facts: ~2 min
Time to load next 100k facts: ~4 min
This trend continues, going from 600k to 700k facts takes over 7
minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux
and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using
large pages, so no memory paging is happening. JRockit is started w/
the -XXagressive parameter, which enables large pages and the more
efficient hash function in HashMap which was introduced in Java5 update
8.
http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html
The end state is over 700k facts, with the possibility of nearly 1M
facts in production. After end state is reached and we issue a few GC
requests, if looks like our memory per fact is almost 9k, which seems
quite high as most of the facts are very simple. Could that be due to
our liberal use of insertLogical and TMS?
We've tried performing a commit every few hundred fact insertions by
issuing a fireAllRules periodically, and that seems to have helped
marginally.
I tried disabling shadow proxies and a few of our ~390 test cases fail
and one loops indefinitely. I'm pretty sure we could fix those, but
don't want to bother if this isn't a realistic solution.
Any thoughts?
Thanks
Eric
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the number
of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of inserted
facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
StatefulSession wm = ruleBase.newStatefulSession();
// Global - output
warnings = new Alarm();
wm.setGlobal(warnings, warnings);
// Add facts
st = (new Date()).getTime();
for(i=0; i nfacts; i++) {
wm.insert(new Data(rand.nextInt(4),
rand.nextDouble()-0.5));
}
en = (new Date()).getTime();
System.out.println(facts = + (en-st));
// Now run the rules
st = (new Date()).getTime();
wm.fireAllRules();
en = (new Date()).getTime();
System.out.println(rules = + (en-st));
// Clean up
wm.dispose();
System.out.println(\n);
}
This code is based on the HelloWorldExample.java code from the manual
and the setup for the rule base is the same as in the manual. As the
number of facts increases runtime increases dramatically:
facts -- runtime (ms)
10 -- 168
20 -- 166
30 -- 344
40 -- 587
50 -- 1215
60 -- 1931
70 -- 2262
80 -- 3000
90 -- 4754
with a maximum memory use of about 428 MB RAM. By contrast, if I use
sequential stateless sessions, everything runs in about 1-5 ms.
Is there something in my set up that would cause this, or is this how
one would expect Drools to scale? I read about people using thousands
of facts so I suspect I'm
RE: [rules-users] Drools 4 poor performance scaling?
I'm in IRC now. The non-business sensitive test case hasn't been maintained. At this stage it might be pretty difficult to create one that doesn't have proprietary information and still functions anywhere the same. We've got nearly 200 rules and 20 different kinds of facts. I wonder if a simple obfuscation would be sufficient? I did give 5.0M1 a try last week. Several of our rules wouldn't compile. I tried for a day or so to fix things, but then gave up. We know it is non-optimal, but we have a few rules with if statements in the RHS and those simply wouldn't compile in 5.0. I'd like to refactor those out to at least an eval in the LHS, but ideally I'd like to precompute the statement and store the result in a new fact so that it could be indexed. Is 5.0 better for multi-threaded access as we discussed before? We've had to wrap all access to working memory in synchronized blocks when using 4.x. That's a pretty big hammer, but it works. Otherwise fact insertions/retracts, firing of rules and queries end up getting run at the same time by different threads and working memory ends up completely unusable. Maybe I'll take another stab at fixing those rules and give 5.0 another go. Any target on a 5.0 release date? We're looking to go live in production in about 1 month. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Mark Proctor Sent: Monday, June 30, 2008 12:39 PM To: Rules Users List Subject: Re: [rules-users] Drools 4 poor performance scaling? Fenderbosch, Eric wrote: We are having a similar problem, although our fact count is much higher. Performance seems pretty good and consistent until about 400k facts, then performance degrades significantly. Part of the degradation is from bigger and more frequent GCs, but not all of it. If you have multi-cpu there is a JVM command you can set a dedicated cpu for GC, that helps somewhat. Time to load first 100k facts: ~1 min Time to load next 100k facts: ~1 min Time to load next 100k facts: ~2 min Time to load next 100k facts: ~4 min This trend continues, going from 600k to 700k facts takes over 7 minutes. We're running 4.0.7 on a 4 CPU box with 12 GB, 64 bit RH Linux and 64 bit JRockit 5. We've allocated a 9 GB heap for the VM using large pages, so no memory paging is happening. JRockit is started w/ the -XXagressive parameter, which enables large pages and the more efficient hash function in HashMap which was introduced in Java5 update 8. Other than the CPU thing, Drools won't take advantage of multipe cpus at the moment. http://e-docs.bea.com/jrockit/jrdocs/refman/optionXX.html The end state is over 700k facts, with the possibility of nearly 1M facts in production. After end state is reached and we issue a few GC requests, if looks like our memory per fact is almost 9k, which seems quite high as most of the facts are very simple. Could that be due to our liberal use of insertLogical and TMS? It could be related to this, especially if you create a long chain of logical relationships. We've tried performing a commit every few hundred fact insertions by issuing a fireAllRules periodically, and that seems to have helped marginally. I tried disabling shadow proxies and a few of our ~390 test cases fail and one loops indefinitely. I'm pretty sure we could fix those, but don't want to bother if this isn't a realistic solution. Any thoughts? Have you tried this on Drools 5.0? It 'doesn't need shadow proxies and implements a new Rete algorithm that is faster for retracts. You can get a nightly build from here, I'd be interested to find out how broken 5.0 is :) https://hudson.jboss.org/hudson/job/drools/lastSuccessfulBuild/artifact/ trunk/target/ We still have more performnace work to do, the items are known, just a matter of time, not all will make 5.0 though. but the main items include: 1) bytecode compiled Rete network, instead of interpreted nodes. I'm hoping this will have a large impact, reducing GC and general indirection and recursive method call frames. 2) true modify, instead of a retract+assert, will also remove the need for activation normalistaion that we do for TMS and the agenda event model. 3) range indexing (initially literals, but would like to explore variables too). Steve, before he left fedex, was creating a simulator for this use case, but removing anything business sensitive. So that we could use it publicly as a benchmark and to help us tune the engine. Are you still working on this? Steve use to chat to us on irc, can I ask you to pop on for a chat? http://labs.jboss.org/drools/irc.html mark Thanks Eric -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel Sent: Thursday, June 26, 2008 12:47 PM To: rules-users@lists.jboss.org Subject: [rules-users] Drools 4 poor performance scaling? I am testing Drools 4 for our application and while sequential mode
RE: [rules-users] Drools 4 poor performance scaling?
A small update. I ran the example below starting with 90 rules and running
down to 1. The execution time remains the same so I don't think the problem
has to do with something left over from the previous call to fireAllRules().
This really should be a simple example and I am still at a loss as to why it
doesn't work quickly.
Again, any help appreciated!
Ron
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ron Kneusel
Sent: Thursday, June 26, 2008 12:47 PM
To: rules-users@lists.jboss.org
Subject: [rules-users] Drools 4 poor performance scaling?
I am testing Drools 4 for our application and while sequential mode is
very fast I get very poor scaling when I increase the number of facts
for stateful or stateless sessions. I want to make sure I'm not doing
something foolish before deciding on whether or not to use Drools
because from what I am reading online it should be fast with the
number of facts I have.
The scenario: I have 1000 rules in a DRL file. They are all of the
form:
rule rule
when
Data(type == 0, value 0.185264);
Data(type == 3, value 0.198202);
then
insert(new AlarmRaised(0));
warnings.setAlarm(0, true);
end
where the ranges checked on the values and the types are randomly
generated. Then, I create a Stateful session and run in a loop timing
how long it takes the engine to fire all rules as the number of
inserted facts increases:
// Run
for(j=0; j 100; j+=5) {
if (j==0) {
nfacts = 1;
} else {
nfacts = j;
}
System.out.println(nfacts + :);
// Get a working memory
StatefulSession wm = ruleBase.newStatefulSession();
// Global - output
warnings = new Alarm();
wm.setGlobal(warnings, warnings);
// Add facts
st = (new Date()).getTime();
for(i=0; i nfacts; i++) {
wm.insert(new Data(rand.nextInt(4),
rand.nextDouble()-0.5));
}
en = (new Date()).getTime();
System.out.println(facts = + (en-st));
// Now run the rules
st = (new Date()).getTime();
wm.fireAllRules();
en = (new Date()).getTime();
System.out.println(rules = + (en-st));
// Clean up
wm.dispose();
System.out.println(\n);
}
This code is based on the HelloWorldExample.java code from the manual
and the setup for the rule base is the same as in the manual. As the
number of facts increases runtime increases dramatically:
facts -- runtime (ms)
10 -- 168
20 -- 166
30 -- 344
40 -- 587
50 -- 1215
60 -- 1931
70 -- 2262
80 -- 3000
90 -- 4754
with a maximum memory use of about 428 MB RAM. By contrast, if I use
sequential stateless sessions, everything runs in about 1-5 ms.
Is there something in my set up that would cause this, or is this how
one would expect Drools to scale? I read about people using thousands
of facts so I suspect I'm setting something up incorrectly.
Any help appreciated!
Ron
_
Need to know now? Get instant answers with Windows Live Messenger.
http://www.windowslive.com/messenger/connect_your_way.html?ocid=TXT_TAGLM_WL_Refresh_messenger_062008
___
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
Re: [rules-users] Drools 4 poor performance scaling?
Ron Kneusel wrote: Mark- I sent the zip file from a separate account because of Hotmail limitations. Please let me know if you got it and thanks for taking a look. Yes I got it, thanks. It will take me a few days to get round to looking into it though, I'll try and get back to you by tue/wed. Ron Date: Thu, 26 Jun 2008 18:37:26 +0100 From: [EMAIL PROTECTED] To: rules-users@lists.jboss.org Subject: Re: [rules-users] Drools 4 poor performance scaling? Ron, Can you send me your project in zip, i'd be interesting to see this. Mark _ Need to know now? Get instant answers with Windows Live Messenger. http://www.windowslive.com/messenger/connect_your_way.html?ocid=TXT_TAGLM_WL_Refresh_messenger_062008 ___ rules-users mailing list rules-users@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users ___ rules-users mailing list rules-users@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users
